Skip to content

Instantly share code, notes, and snippets.

@jakirkham

jakirkham/cupy_chain_conv_test.ipynb

Created September 11, 2019 17:18
Show Gist options

  • Select an option

    Learn more about clone URLs
  • Save jakirkham/2c66b096d472054ccb1c9556bad9601a to your computer and use it in GitHub Desktop.

Select an option

Learn more about clone URLs
Save jakirkham/2c66b096d472054ccb1c9556bad9601a to your computer and use it in GitHub Desktop.
Download ZIP
Attempt at using CuPy + Chainer to mimic SciPy's convolve
Display the source blob
Display the rendered blob
Raw
cupy_chain_conv_test.ipynb
{
"cells": [
{
"cell_type": "code",
"execution_count": 1,
"metadata": {},
"outputs": [],
"source": [
"import numpy as np\n",
"import scipy as sp\n",
"import scipy.ndimage as spimg\n",
"import cupy as cp\n",
"import chainer\n",
"from chainer.functions import convolution_nd"
]
},
{
"cell_type": "code",
"execution_count": 2,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([[0.55021381, 0.60364142, 0.41514555, ..., 0.21014848, 0.06411803,\n",
" 0.00318856],\n",
" [0.565308 , 0.70801313, 0.57825412, ..., 0.77520092, 0.23019033,\n",
" 0.94941756],\n",
" [0.94411044, 0.75518582, 0.3173873 , ..., 0.98212996, 0.45765117,\n",
" 0.46621948],\n",
" ...,\n",
" [0.11534148, 0.09391876, 0.88002047, ..., 0.21290276, 0.96188936,\n",
" 0.75469962],\n",
" [0.69817622, 0.34021495, 0.88724349, ..., 0.96765835, 0.90653356,\n",
" 0.56921228],\n",
" [0.57407528, 0.7842093 , 0.08633729, ..., 0.60257705, 0.96884379,\n",
" 0.37288173]])"
]
},
"execution_count": 2,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"cp.random.seed(42)\n",
"a = cp.random.random((100, 110))\n",
"a"
]
},
{
"cell_type": "code",
"execution_count": 3,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([[0. , 0. , 0. , ..., 0. , 0. ,\n",
" 0. ],\n",
" [0. , 0.55021381, 0.60364142, ..., 0.06411803, 0.00318856,\n",
" 0. ],\n",
" [0. , 0.565308 , 0.70801313, ..., 0.23019033, 0.94941756,\n",
" 0. ],\n",
" ...,\n",
" [0. , 0.69817622, 0.34021495, ..., 0.90653356, 0.56921228,\n",
" 0. ],\n",
" [0. , 0.57407528, 0.7842093 , ..., 0.96884379, 0.37288173,\n",
" 0. ],\n",
" [0. , 0. , 0. , ..., 0. , 0. ,\n",
" 0. ]])"
]
},
"execution_count": 3,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"ap = cp.pad(a, 1, \"constant\", constant_values=0)\n",
"ap"
]
},
{
"cell_type": "code",
"execution_count": 4,
"metadata": {},
"outputs": [],
"source": [
"f = cp.ones(a.ndim * (3,))"
]
},
{
"cell_type": "code",
"execution_count": 5,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([[2.42717635, 3.42057602, 3.0604331 , ..., 2.49181719, 2.23226387,\n",
" 1.24691447],\n",
" [4.12647261, 5.43725958, 4.61147945, ..., 4.29905919, 4.13826448,\n",
" 2.17078512],\n",
" [4.47953769, 5.9480567 , 5.47944319, ..., 5.0753601 , 5.58448565,\n",
" 3.27832063],\n",
" ...,\n",
" [2.83098768, 4.73980507, 3.66308175, ..., 5.2031273 , 6.30222493,\n",
" 4.35831288],\n",
" [2.605936 , 4.45953725, 3.90981577, ..., 6.08762965, 6.3171985 ,\n",
" 4.53406034],\n",
" [2.39667576, 3.37025653, 2.85051356, ..., 4.85585691, 4.38770676,\n",
" 2.81747136]])"
]
},
"execution_count": 5,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"r = convolution_nd(ap[None, None], f[None, None]).array[0, 0]\n",
"r"
]
},
{
"cell_type": "code",
"execution_count": 6,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([[2.42717635, 3.42057602, 3.0604331 , ..., 2.49181719, 2.23226387,\n",
" 1.24691447],\n",
" [4.12647261, 5.43725958, 4.61147945, ..., 4.29905919, 4.13826448,\n",
" 2.17078512],\n",
" [4.47953769, 5.9480567 , 5.47944319, ..., 5.0753601 , 5.58448565,\n",
" 3.27832063],\n",
" ...,\n",
" [2.83098768, 4.73980507, 3.66308175, ..., 5.2031273 , 6.30222493,\n",
" 4.35831288],\n",
" [2.605936 , 4.45953725, 3.90981577, ..., 6.08762965, 6.3171985 ,\n",
" 4.53406034],\n",
" [2.39667576, 3.37025653, 2.85051356, ..., 4.85585691, 4.38770676,\n",
" 2.81747136]])"
]
},
"execution_count": 6,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"e = spimg.convolve(cp.asnumpy(ap), cp.asnumpy(f))[1:-1, 1:-1]\n",
"e"
]
},
{
"cell_type": "code",
"execution_count": 7,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"True"
]
},
"execution_count": 7,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"np.allclose(cp.asnumpy(r), e)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.7.3"
}
},
"nbformat": 4,
"nbformat_minor": 4
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment